Crappy CDIL datasheet...

[Piotr Wyderski]

Here is an interesting part for my application, especially because its
very low C_in:

https://www.tme.eu/Document/901eddfa360c5ef73460d5ceeadb1a22/CDD20N03.pdf

But I have no clue how to parse this datasheet. On one hand, they say
the maximum V_GS is 12V, which is OK and suggest a logic-level device.
On the other, most of the specs are at V_GS=10V, which is scary if 12V
is the abs. max. Figure 9 at page 4, the R_DS(V_GS) seems to come from
another part, as they claim the fraction of a milliohm range. At the
first page they claim R_DS_ON < 35mOhm@10V. It looks like a random
collection of pages.

Could someone more experienced with this type of \"documentation\" tell me
what R_DS_ON can I expect at V_GS=5V?

Best regards, Piotr

 

[Anthony William Sloman]

On Sunday, March 27, 2022 at 7:00:07 PM UTC+11, Piotr Wyderski wrote:

Here is an interesting part for my application, especially because its
very low C_in:

https://www.tme.eu/Document/901eddfa360c5ef73460d5ceeadb1a22/CDD20N03.pdf

But I have no clue how to parse this datasheet. On one hand, they say
the maximum V_GS is 12V, which is OK and suggest a logic-level device.
On the other, most of the specs are at V_GS=10V, which is scary if 12V
is the abs. max.

12V Gate-to-source is the the voltage they guarantee that the part will survive. As John Larkin points out from time to time, testing where the gate breaks down would destroy the part, so they don\'t do it, and you can probably get away with applying more volts to most parts.

Figure 9 at page 4, the R_DS(V_GS) seems to come from
another part, as they claim the fraction of a milliohm range. At the
first page they claim R_DS_ON < 35mOhm@10V. It looks like a random
collection of pages.

The number on the vertical axis are probably actually in ohm, but some clown apparently decided to label them as milliohms.

Could someone more experienced with this type of \"documentation\" tell me
what R_DS_ON can I expect at V_GS=5V?

At low current, perhaps 0.08 ohm. It seems to get up to 0.24 ohm at 9A, but note Fig 5, where the drain current hits a limit of about 7A at VGS=4V.

It is spelled out in Figure 6, plotted as a function of drain current for both 4.5V gate-to-source and 10V .

--
Bill Sloman, Sydney

 

[server]

On Sun, 27 Mar 2022 09:59:47 +0200, Piotr Wyderski
<bombald@protonmail.com> wrote:

Here is an interesting part for my application, especially because its
very low C_in:

https://www.tme.eu/Document/901eddfa360c5ef73460d5ceeadb1a22/CDD20N03.pdf

But I have no clue how to parse this datasheet. On one hand, they say
the maximum V_GS is 12V, which is OK and suggest a logic-level device.

Actually, it doesn\'t. It\'s a fairly high threshold part.

On the other, most of the specs are at V_GS=10V, which is scary if 12V
is the abs. max.

10v is not scary, and parts are not supposed to fail at abs max. To
get real performance, parts sometimes need to be run at abs max, or
beyond. Abs max specs are often ultra conservative; I test parts to
failure and back off some.

Mosfet gates typically fail around 60 volts or so. There could be
long-term degradation a bit below that.

Buy a few and test them!

Figure 9 at page 4, the R_DS(V_GS) seems to come from

another part, as they claim the fraction of a milliohm range. At the
first page they claim R_DS_ON < 35mOhm@10V. It looks like a random
collection of pages.

The initial slopes of fig 5 indicate typical Ron.
Fig 9 looks about right if the units are ohms, not mohms.

Could someone more experienced with this type of \"documentation\" tell me
what R_DS_ON can I expect at V_GS=5V?

Best regards, Piotr

The state of data sheets in the semi business is a disgrace. They must
assign the dullest, least numerate interns to composing data sheets,
and seem to never correct even gross errors.

Someone could do a fun web site pointing out bad data sheets, and
buggy parts.

RF part data sheets are invariably bad.





--

I yam what I yam - Popeye

 

[Piotr Wyderski]

jlarkin@highlandsniptechnology.com wrote:

> Actually, it doesn\'t. It\'s a fairly high threshold part.

This is true, but then who in their right mind specifies 12V max and
expects it to be driven at 10?

> 10v is not scary, and parts are not supposed to fail at abs max.

If the design is supposed to stay within the abs max, it leaves you only
2V of margin for ringing. Regular FETs specify +/-20V for 10V operation.

To get real performance, parts sometimes need to be run at abs max, or
beyond. Abs max specs are often ultra conservative; I test parts to
failure and back off some.

Indeed, but in the first place the part must be inherently capable of
some \"real performance\". This is a boring 30V FET. Its two advantages
are low Ciss and super-low price. I need many and the 8mA available from
a 74HCT595\'s output would still allow substantially sub-microsecond
switching speeds, tremendously simplifying the module.

> Buy a few and test them!

Not sure if it is the right approach. The next reel might have different
specs. Nah, I think I will pay 2x its price to get a part with reliable
specs. The AOD454, for instance.

The initial slopes of fig 5 indicate typical Ron.
Fig 9 looks about right if the units are ohms, not mohms.

This was my initial assumption, but then the chart says sort of 80mOhms
at 10V, while the first page says \"RDS(ON) <35mΩ @ VGS=10V\". This is 2x off.

Best regards, Piotr

 

[server]

On Sun, 27 Mar 2022 18:14:51 +0200, Piotr Wyderski
<bombald@protonmail.com> wrote:

jlarkin@highlandsniptechnology.com wrote:

Actually, it doesn\'t. It\'s a fairly high threshold part.

This is true, but then who in their right mind specifies 12V max and
expects it to be driven at 10?

What\'s wrong with that? Seems perfectly sensible to me. 10 is less
than 12.

10v is not scary, and parts are not supposed to fail at abs max.

If the design is supposed to stay within the abs max, it leaves you only
2V of margin for ringing. Regular FETs specify +/-20V for 10V operation.

To get real performance, parts sometimes need to be run at abs max, or
beyond. Abs max specs are often ultra conservative; I test parts to
failure and back off some.

Indeed, but in the first place the part must be inherently capable of
some \"real performance\". This is a boring 30V FET. Its two advantages
are low Ciss and super-low price. I need many and the 8mA available from
a 74HCT595\'s output would still allow substantially sub-microsecond
switching speeds, tremendously simplifying the module.

An HCT595 is rated 7 volts abs max. Nowhere near 12.

And it will source a lot more than 8 mA, even if you only use one
section.

Buy a few and test them!

Not sure if it is the right approach. The next reel might have different
specs. Nah, I think I will pay 2x its price to get a part with reliable
specs. The AOD454, for instance.

The initial slopes of fig 5 indicate typical Ron.
Fig 9 looks about right if the units are ohms, not mohms.

This was my initial assumption, but then the chart says sort of 80mOhms
at 10V, while the first page says \"RDS(ON) <35m? @ VGS=10V\". This is 2x off.

Best regards, Piotr

Better buy something else.



--

I yam what I yam - Popeye

 

[server]

On Sun, 27 Mar 2022 09:59:47 +0200, Piotr Wyderski
<bombald@protonmail.com> wrote:

Here is an interesting part for my application, especially because its
very low C_in:

https://www.tme.eu/Document/901eddfa360c5ef73460d5ceeadb1a22/CDD20N03.pdf

But I have no clue how to parse this datasheet. On one hand, they say
the maximum V_GS is 12V, which is OK and suggest a logic-level device.
On the other, most of the specs are at V_GS=10V, which is scary if 12V
is the abs. max. Figure 9 at page 4, the R_DS(V_GS) seems to come from
another part, as they claim the fraction of a milliohm range. At the
first page they claim R_DS_ON < 35mOhm@10V. It looks like a random
collection of pages.

Could someone more experienced with this type of \"documentation\" tell me
what R_DS_ON can I expect at V_GS=5V?

Best regards, Piotr

Take a look at TPIC6595.



--

I yam what I yam - Popeye

 

[Piotr Wyderski]

jlarkin@highlandsniptechnology.com wrote:

What\'s wrong with that? Seems perfectly sensible to me. 10 is less
than 12.

If driven gently -- nothing wrong. Pumping a lot of current abruptly
will cause overshoots, though. 20% safety margin is far less than the
typical 100%. I wouldn\'t consider the part in a 10V system for this
reason alone. However, this is a 5V system, so the margin is still huge,
even at 12V. No issue here. But the specs don\'t make sense at this V_GS,
so I will better keep away from this part.

> An HCT595 is rated 7 volts abs max. Nowhere near 12.

The resistance in the 5-7V range is unknown. Even at the rated 10V you
get contradicting numbers, depending on what chart/page you like more.

And it will source a lot more than 8 mA, even if you only use one
section.

70mA total abs max and I want to use all the channels. That makes it
8*8=64mA + some margin. Still, a microsecond is good enough.

> Better buy something else.

Yes, the AOD454 looks decent and the datasheet is not
self-contradictory. Above all, it is fully specified at 4.5V.

Best regards, Piotr

 

[Piotr Wyderski]

jlarkin@highlandsniptechnology.com wrote:

> Take a look at TPIC6595.

I absolutely love this part, or, to be specific, its Nexperia\'s sister
due to better specs. I am pulsing the -OE at 1MHz rate and drive 8
flyback transformers directly from the chip. But it is open-drain, there
is no HV push-pull version as far as I know. The FET drivers would be
more complex than they need to be. Hence the 595 to close the gap: the
same digital interface, 7V-capable and push-pull. The 70mA total is a
bit low, but there are low Qg FETs and then the FET driver turns out to
be a 560 Ohm gate resistor. Cheap and reliable.

Best regards, Piotr

 

[server]

On Sun, 27 Mar 2022 20:26:29 +0200, Piotr Wyderski
<bombald@protonmail.com> wrote:

jlarkin@highlandsniptechnology.com wrote:

What\'s wrong with that? Seems perfectly sensible to me. 10 is less
than 12.

If driven gently -- nothing wrong. Pumping a lot of current abruptly
will cause overshoots, though. 20% safety margin is far less than the
typical 100%. I wouldn\'t consider the part in a 10V system for this
reason alone. However, this is a 5V system, so the margin is still huge,
even at 12V. No issue here. But the specs don\'t make sense at this V_GS,
so I will better keep away from this part.

An HCT595 is rated 7 volts abs max. Nowhere near 12.

The resistance in the 5-7V range is unknown.

The HCT output resistance? Measure it. These parts are not specified
as mosfet gate drivers, so if you want to use them, experiment.

Even at the rated 10V you
>get contradicting numbers, depending on what chart/page you like more.

The data sheet curves for the fet are pretty complete. Derate a bit
maybe for part variations. Again, measure to confirm.

And it will source a lot more than 8 mA, even if you only use one
section.

70mA total abs max and I want to use all the channels. That makes it
8*8=64mA + some margin. Still, a microsecond is good enough.

Charging and discharging some mosfet gates won\'t overheat an HCT part.
Or cause electromigration.

Better buy something else.

Yes, the AOD454 looks decent and the datasheet is not
self-contradictory. Above all, it is fully specified at 4.5V.

Best regards, Piotr

--

I yam what I yam - Popeye

 

[Anthony William Sloman]

On Monday, March 28, 2022 at 3:15:09 AM UTC+11, Piotr Wyderski wrote:
> jla...@highlandsniptechnology.com wrote:

<snip>

Fig 9 looks about right if the units are ohms, not mohms.

This was my initial assumption, but then the chart says sort of 80mOhms
at 10V, while the first page says \"RDS(ON) <35mΩ @ VGS=10V\". This is 2x off.

Figure 9 is at a drain current of 3A. Figure 6 shows a significant rise in drain resistance with increasing drain current.

Figures are mostly typical values. The first page number is talking about the worst cast drain resistance. They are different. Three-to-one differences between best case and worst case numbers aren\'t unusual.

--
Bill Sloman, Sydney